Is that possible apply WaNet attack on ImageNet?

[chengyi-chris]

Thank you for your impressive work. I would like to ask for details on how to apply WaNet to the ImageNet dataset (similar to the CIFAR10 setting). How can I modify the code accordingly? The code create_poisoned_set_imagenet.py generates the following error message:

'''
File "/work/u1056911/backdoor-toolbox/poison_tool_box/WaNet.py", line 127, in transform
data = F.grid_sample(data, grid_temps.repeat(data.shape[0], 1, 1, 1), align_corners=True)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/u1056911/.local/lib/python3.11/site-packages/torch/nn/functional.py", line 4304, in grid_sample
return torch.grid_sampler(input, grid, mode_enum, padding_mode_enum, align_corners)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
RuntimeError: grid_sampler(): expected grid to have size 1 in last dimension, but got grid with sizes [3, 224, 224, 2]
'''
Do you know what the problem might be?